Cisco has about $5 billion in cash & short-term investments. Nokia's market cap is $71 billion (and would probably need an offer that values the firm at $80 to $90 billlion to succeed). Even if Cisco liquidates its $11 billion in long-term investments, it can't swing this deal with what it has on hand.
Unless Cisco goes into major debt with a leveraged buyout, they can't afford to buy someone as big as Nokia.
What about creating an ad hoc distributed network of sensors (versus a fixed network). If thousands or millions of people downloaded a worm monitor application, then the sensor network would be very fluid and span IP space in a less predictable way. An ongoing P2P cross-comparison of the signatures of unsolicited packets could also provide distributed detection of novel worms. When too many sensors see the same anomalous thing, the alert propagates across the network.
Done well, it would create an internet immune system in that sensors that had seen the worm would alert machines that hadn't seen the worm. Those machines would then automatically filter for the new pattern while watching for confirming evidence that a worm was loose. If the download also provided a protective feature, then more people would download it and that the network would become more sensitive and valuable.
The idea may have some minor problems. First, security vulnerabilities could be introduced by the monitoring package itself (e.g, the Witty worm targeted vulnerable firewall software). I'd recommend that if a buffer overflow or malformed command exploit were ever discovered in Honeypot@home code, then all the developers would have to be shot immediately. Second, I see some but no great problem of worm writers trying to subvert the network because it would be hard for them to register enough machines (and replace the code to mute the alert signal triggered by their worm) to swamp the alert signals generated by legitimate nodes of the network.
There is a large assumption here, which is that only the customer who was issued the card is the person that ever uses the card, which is not true. Especially when said cards have incentives like "points" that can be accumulated to obtain free items, or a straight discount for using the card. Such card "borrowing" will skew the data in the database, since the computer assumes that the card equals the person, which is not necessarily true.
Card borrowing does skew the data, but how high is the actual frequency of this practice? Moreover, I'd bet that many card borrowings are between demographically similar people. If one 22 year old college student borrows a card from another 22 year old college student, then the effect on Tesco is minimal. Yes, card borrowing affects the data, but I doubt it ruins it.
You could argue that Tesco's buying policy really is only concerned with the aggregate information anyway since they won't make a purchase decision based on a single cardholder, and this aggregate is unchanged by card "swapping". But then why do you need to track individual information?
Aggregate data only gets a company so far. Consider the problem deciding where to put a new store. The only way Tesco can accurately do this is by cross comparing the census records for the proposed area with Tesco's own data on who buys how much and from what distances at its existing stores. If Tesco knows that people drive 20 miles to reach a Tesco, then its less likely to put another Tesco within 10 miles of an existing one. And if Tesco sees certain demographic groups is a new area, they can estimate likely sales and the best product assortment by studying the sales habits of those demographic groups within its loyalty card program. Breaking down the customer base by distance to store, types of goods bought, ages, income, etc. all help the company do a much better job of store siting, merchandise assortment, forecasting, etc.
Yes, only aggregate data is useful. But the loyalty card program lets Tesco aggregate data in new ways -- aggregate by customer age, gender, distance to store, household income, or tendencies to buy products. Store-level aggregate data or even simple market basket analysis isn't going to give as good an insight into the relationships between customer categories and shopping habits.
The answer is because they can. However the whole concept of "keeping a file" on someone is rather macabre. I have access to my criminal record, my medical record and my credit record. I should have access to all information ANYONE keeps on me unless it's being used to prosecute me for a crime. When is this going to happen at the supermarket? Until then, I refuse to use these cards.
I very seriously doubt that Tesco spends millions of pounds (probably tens of millions of pounds) on its loyalty card database just because it can. Most businesses insistent on a bottom-line improvement in the business before laying out that kind of money. And if a company does spend that type of money and doesn't get a profitable return, they stop the program. Morevoer, the type of data collected by a loyalty card program is perfect for assessing the business benefits of a loyalty card program. Tesco started their program 10 years ago. I doubt it would still be in use today if it didn't provide direct benefits to the company.
In essence, isn't the Tesco loyalty card system like a sophisticated representative democratic process? People "register to vote" using a loyalty card, vote by buying goods, and Tesco watches the results of the "election." Tesco knows who buys what and can thus go to suppliers and argue for changes that are more likely to satisfy customers.
Although non-loyalty card users still count at Tesco (all retail is a type of democracy in that people vote with their pocketbooks), I'm sure that the product choices of loyalty card users are far more influential with Tesco and thus with suppliers. In that regard not having a loyalty card is like not having a voter registration card.
Some might argue that voting should be anonymous, like political democratic elections, and perhaps it should. Yet non-anonymous voting provides valuable information -- e.g., Tesco might notice that, for example, people who buy lots of hot soups in the winter don't buy high-sugar fizzy drinks in the summer or some other correlation that is only observable if you can know that shopping basket A and basket B (6 months later) represent the same voter. These long-term correlations aid in both store assortment planning and forecasting.
It's easily the worst airport in the entire United States of America
Agreed! Trying to find a working electrical outlet in Concourse C (at least) is a true exercise in futility. Apparently MassPort disabled nearly all of them for some pathetic reason (concession to the for-pay internet kiosk maybe?).
GRRR! This thread reminded me that I have to fly there twice this month.
While I applaud attempts to secure WiFi, it would seem that wireless will always add another channel of vulnerability to any IT system, especially because WiFi is so often deployed inside the firewall. WiFi system are generally vulnerable to both internet-based attacks and wireless attacks. And even if the 802.11i protocol "secure," there is little guarantee that both the AP and the client wifi transceiver have a secure implementation of the protocol or that the user configures the system in a secure fashion.
As inconvenient as wires are (and even they are not totally secure), they do reduce the amount of one personal information freely broadcast into the ether.
You raise an excellent argument but I fear that it creates the opposite of what you want. Credit information is far more valuable (to humanity and civilization) than are the latest music files by Brittany Spears.
If you study banking in China you find that one of the big problems over there is a lack of credit information systems. Its easy for someone to get a loan, skip the payments, go get another loan at another bank, skip the payments, and repeat as needed. In such a system honest people pay the price (high interest rates) to cover losses generated by dishonest borrowers. Without some mechanism for sharing credit histories, its almost impossible to have a viable credit card system or low-cost consumer loans (I'll leave it to others to argue whether these are Good Things or not).
The problem, and it is a massive one, is not that people are collecting the information, but that they are misusing the information or allowing to be misused by failing to secure it against criminal incursions. The same aggressive defense that prevents counterfeit currency in the U.S. should be applied to those that would counterfeit identities with stolen information. Your point about Choicepoint is well taken -- collectors of personal financial information should be held very accountable (and liable) for lapses in their security and for the actions of those they give data to.
your "recomputing/rerendering the scene parallax in realtime" betrays your having forgotten that we're talking about film here...
Film records a fixed perspective but human heads and eyes are not fixed. And that's the problem with film for use in 3D. FIlm is not an appropriate medium for visually comfortable 3-D because it forces the viewer to hold their head in the same fixed orientation used by the cameras.
The core challenge for 3-D is creating a system that works when a person tilts their head. Current 3-D filming and multi-person viewing systems assume that the viewers left eye is a fixed horizontal distance to the left of the right eye with no vertical displacement between the eyes' pupils. This assumption is only true when everyone is sitting upright in their chairs. If the viewer tilts their head, then the parallax of the scene appears unnaturally displaced and gives the viewer eyestrain, headache, or a sensation of double-images. With 3-D, you can't rest your head on your partner's shoulder, tilt your head to see around the person in front of your, or lie on the couch and watch it without some visual discomfort. I'd imagine that most people won't consciously notice the problem but might subconsciously become aware that they get eyestrain, neck-pains, headaches, or a vaguely nauseous disoriented feeling when they see a 3D movie -- not a recipe for repeat business.
One nearterm solution to the problem is constructing tilt-dependent parallax for each viewer. The person with their head tilted to the right needs to see a different pair of images than the person who is sitting up straight or who has tilted their head to the left. This pushes 3D into the realm of more awkward and more expensive personal viewing headsets and the need for tracking head tilt and recomputing/rerendering the scene parallax in realtime.
The longterm solution is holographic or volumetric systems that create/reconstruct an optical 3-D field. This solves the head tilt problem, although adds the minor cinematic problem that the people on the left side of the theatre may have an obstructed view (relative to the people in the center or right-side) if, for example, the main character's hand covers some important object from some angles.
There's an old story from the early neural net image recognition days that seems germane to this. A group of researcher were trying to train an artificial neural net to recognize military tanks that were partially hidden in forested scenes (this was the bad old Cold War days and spotting Soviet tanks in West German forests was the problem du jour). Pictures of natural forested scenes with and without tanks were used to train and test the system. It seemed to work very well on all the training and test data.
But when they tried the system on more images, it failed miserably. Further investigation revealed that, by accident, all of the "tank" pictures had been taken on cloudy days and all of the non-tank pictures had been taken on sunny days. The system had learned, and learned beautifully, how to recognize cloudy vs. sunny days.
The point is that the software was good enough to learn to recognize the difference between the two populations of images but that that difference wasn't the one intended by the people working on the system. In the same vein, I'm sure that Peekaboom will learn to distinguish between objects in images but whether it learns the actual object or just some incidental characteristic of that pocture of the object will require a very very good diversity of training pictures to avoid accidental, non-meaningful patterns in the image data.
I do wish them luck. Perhaps Peekaboom could create a distributed version of the training process in which others can both submit and help train on new objects/images. Letting others submit images and train the system would help diversify the training & testing data sets. Because some people will, no doubt, submit porn, I'm sure the system might become quite adept at recognizing the nether regions of the human body.
Technological anonymity does not address all the vulnerabilities of the system because the identity of the source is also stored in the mind of the journalist. In fact, I suspect that technological security is an iron padlock on a paper house -- the human factors/social engineering issues create some severe vulnerabilities in the system. If the government can threaten or actually imprison a journalist over sources, then encrypted HDs aren't going to be much of a defense. As long as the journalist can ID the source, the system is very vulnerable.
The problem is that journalists really can't use unverified sources -- it's to easy to be either wrong, manipulated, or both as Dan Rather can attest to.
The real trick would be to accomplish both credibility and anonymity. Somehow the journalist needs to both know the source well enough to be sure that they are credible and yet not know the source well enough to ID them. Then they can add whatever encryption/obfuscation they want because none of the vulnerable information lies in a human mind.
If the OSS community could convince AOL to add a Linux distro to their ubiquitous CDs, I'm sure it would reach a lot of people. AOL may not be blanketing the world with disks like they used to, but they are still everywhere (in USPS change-of-address packets, next to store cash registers, and in the occasional Sunday newspaper).
I've never used it (maybe it deserves to die) but I'm surprised IBM didn't spin-off OS/2 sales & support as a little services company (with an appropriate slice of the proceeds of the service contracts). If people want to use OS/2, why not sell it to them? If people need support for it, why not sell it to them?
I could understand a company killing a product that competes with its own more modern systems, but how do continued OS/2 sales hurt IBM more than orphaning some existing customers?
I'm going to have to both agree with you and disagree with you .
Why is 300 dollars a limit? Warren Buffet's stock costs thousands for a single share! A much more important question is whether Google is worth it's valuation. GOOG is worth about 80 billion. YHOO is worth about 50. So the number isn't wildly off target. Remember, share price is determined in part by how much of the company that share represents, in google's case, it's much thinner.
Absolutely true. Anyone who thinks that some $/share is cheap or expensive without understanding total valuation is not going to last long in the stock market. The bigger challenge is figuring out what a share buys in terms of future returns.
Especially once you look at the rest of the numbers. YHOO isn't expected to grow nearly as fast as GOOG. Last quarter, both brought in about the same revenues, but google had a better EBITDA (Earnings before bs you have to put up with), and Google's revenue vs last quarter grew at 100 percent vs Yahoo's 55. The earnings growth is equally impressive. Something like 400 percent to 100. Sounds like a ripe time for MSFT and other competitors to walk in and make some money, but a difficult time to actually kill off anybody competing for bright people.
Actually, Yahoo's revenues are growing faster than Google's. Earnings growth is a total red herring in fast-growing companies because its the difference of costs and revenues. In a young company its easy for these to flucuate and for the difference to gyrate madly (a zillion percent earning growth may mean nothing is the baseline was near zero). EBITDA is even worse because it subtracts numbers such as taxes, depreciation, and amortization that do come out of the company when it has matured. Because YHOO and GOOG are in much the same industry, they should have similar margins once they stabilize. Yet GOOG price/sales ratio is nearly twice that of YHOO's (21.88 vs. 12.85) suggesting that GOOG's margins need to be twice that of YHOOs to make up the difference. And if you care about it, YHOO has 4 times the cashflow of GOOG (although this number is hard to interpret in a rapidly expanding companies).
Ok, so I think the numbers are great, but what about contrarians? There were plenty of people worried at the IPO that the strike price would decimate initial investors, but the price jumped to like 130. When it reached 200 people suggested it was too high. And now at 300, how many people think this thing is too much? 3.5 percent of the GOOG float (available stock) is shorted. 6.2 percent is shorted in YHOO. I suppose this question is better answered by asking how much money is shorting each, and the answer is 2.84 billion dollars in YHOO, and 2.90 billion short in GOOG. So maybe, unless you consider that about 30 billion extra dollars feel that google is more valuable. But I'd be more concerned that a majority share is held by insiders concerned more about their own interests than the interests of all stockholders.
You are very perceptive to estimate the total cap of shorted stock. But I'd be careful about reading too much into the short ratios because shorters consider more than just the long-term future value of the stock in shorting. GOOG is currently a daring. If it is in a bubble, then shorters will avoid it even if they know that GOOG is horribly overvalued. Short sellers have a fixed maximum gain (i.e., when the stock goes to 0) but unlimited potential loss (e.g., if the stock double, triples, etc. in value). Because bubbles have a nasty habit of being unpredictable -- nobody knows when GOOG will hit the wall -- shorters will avoid GOOG. In contrast, YHOO has been through the dot-bomb ringer and that will make bulls more cautious and the bears more agressive. Short sellers know that YHOO is less likely to be caught up in a prolonged mania phase. The point, both GOOG and YHOO may have
The other front-page story on Netscape highlights the promises and risks of high-flying internet companies. In this post I argue that Netscape fell because it was so easy to switch browsers, especially when getting a new computer.
I wonder if Google will be able to make itself sticky enough to survive any threats? Currently, Google doesn't really offer any intercompatibility advantages in the sense that a co-worker's use of Google does not influence my use of Google. And if I replace my PC, Google doesn't offer anything that encourages me to use Google on the new machine. (GMail is somewhat sticky, but is too independent of Google's core search to force people to stay with Google search)
In contrast, I can see how MS could offer more integrative search experience where people would use MS search tools because friends and coworkers use MS search tools. If my coworker's PC is indexed by MS, my old PC was indexed by MS, and my new PC comes with built-in local/global search tools, then I'd bet a large fraction of people will switch to MS search tool regardless of Google's marketshare. Even Google's ad-words placements on 3rd party sites could be threatened if nex-gen MS server includes integrated ad serving tools.
I hope that Google finds a way to encourage people to stay with Google even as they change PCs or interconnect with co-workers and friends. The current valuation of Google requires both high growth and low risk.
Netscape's rise and fall epitomizes the acceleration of the business cycle. The fact that anyone can download anything at low cost and the fact that most people replace their computers every 2 years means a new, small company can quickly grow its customer base. And those same tools meant that MS could, just as quickly distribute its own browser and quickly take Netscape's installbase from the company.
Low distribution costs and PC turnover means that marketshare leadership is not assailable under most conditions -- its too easy for people to replace old software, especially when they get a new computer. Only companies that have an interoperability hook that ties past, present, and future generations of software and systems together have any hope of retaining marketshare.
MS has tried, and succeeded, in creating that hook with IE in that many websites "work best" with Explorer and Windows-specific web functionality (VBscript, ActiveX, MS-extensions to javascript, etc.). To the extent that MS is forced, in the future, to embrace true open standards (not embrace-and-extent forks of those standards) then the OS and app maker will become vulnerable to rapid changes in marketshare.
It sounds like we need new directives in robots.txt to categorize what's permitted and what's not permitted regarding caching, archiving, display of conextual text around a search hit, time-to-live, copyright jurisdiction, etc.
If some countries want to implement asinine policies, at least sites should have the decency (and the means) to let global information services, such as Google, respect those policies.
Connectivity - global media, the internet -- have created a winner-take-all world that drives both the creators of studies and the reporters of studies toward hyperbole. If someone wants their 15 minutes of fame, they need to do (or appear to do) something spectacular. When attention is a scarce resource (because of an explosion of applications/demands for attention), then it drives people toward excessive behavior in crafting and reporting the results of studies.
At the same time, I wonder if the long tail efect means that an increasing number of once-obscure, high-quality studies are being discovered, read, and used by an increasing number of people. Those that do create unbiased studies may not get much popular press, but they do become more widely read due to Google.
Ultimately, we seem to be floating in a rising tide of both good and bad studies. Perhaps the ratio of studies is being biased toward the bad (winner take all) but the ratio of impressions -- the numbers of times that good studies have been accessed -- has actually improved due to long-tail effects.
If everyone switches to wideband, low-power, densely-coded, mesh-network transmissions, then I suspect that the Earth will become virtually invisible to extraterrestrials who try to use SETI-style, pattern-in-RF methods. With nobody broadcasting at high power on a simple-coded narrow-band carrier, the RF emissions of the planet will become indistinguishable from noise.
I wonder if each civilization goes through a short RF-detectability phase before they so densely pack the spectrum with so many emitters that they become invisible, too.
Rise of software-embodied functionality
on
Software Telescope
·
· Score: 4, Interesting
This development highlights the ongoing replacement of specialized, engineered devices with general purpose CPU + software. So many things (car's carburetors, motor speed controls, printers, appliance controls, radios, etc.) were formerly designed using mechanical and electrical circuits that implemented the needed functionality. Now they do it with a CPU such as an embedded controller and a bit of code in flash-RAM.
The shift from hardware-embodied functionality to software-embodied functionality is very profound because of the differences in cost structures. The cost of complexity for software is far lower than the corresponding cost of complexity for hardware. The cost of manufacturing for software is lower than the cost of manufacturing for hardware. The cost of modifying or upgrading software is far lower than the cost of replacing or upgrading hardware. Products with software-embodied functionality can be designed at low cost, made in volume at lost cost and changed at will after sale. The result is greater variety and faster development of new products.
The effects go much farther than cost and variety. Perhaps the most interesting effect is that caused by Moore's law. Whereas mechanical and electronic system have not improved much in the last few decades, CPUs have. Creating software-embodied systems means that the device's performance can become slaved to Moore's law -- a software-embodied product gets a free performance boost with every doubling of transistor count/clockspeed.
I apologize for using this/. cliche, but I, for one, welcome our new software-embodied overlords.
The low-end has always, eventually, overtaken the high-end. Minicomputer became as powerful as mainframes, Workstations became as powerful as minicomputers, and PCs became as powerful as workstations. There will always be room for some high-end gear, but total marketshare for high-end stuff isn't growing and some of the high-end players have died.
The interesting question is when consumer electronics will replace PCs as the most ubiquitous computing device. They are not there yet, but I'd wager that next generation consoles could easily run most of the applications that most people currently use a PC for (web, email, office, and games). Of course some people will always seek out $2,000+ high-end PCs, but more and more people will opt for a $300-$500 unit like a game console (with disk) or something like a Apple's Mini.
I'm sure this is why MS has devoted so much money to XBox and Apple created the Mini. They learned the lesson (that DEC both epitomized but failed to learn) that the low-end rises and supplants the high-end.
An estimate of the orbital delta-v for Tempel/Deep Impact suggests a velocity change of only 1 cm/hour (I can't vouch for the math). Assuming we would need to nudge a threatening body by 1/2 the diameter of the Earth (from direct hit to grazing pass-by), we would need to know to hit a Tempel 1-sized body in advance by over 73,000 years. This type of mission would work 10 years in advance for much smaller bodies (say less than 350 m in diameter). Even these estimates assume a perfect strike by the deflecting deep impactor -- a margin of error or the need to push the object several Earth-diameters further reduces the potential for this method.
Kinetic energy is not the way to go. Deep Impact delivered only about 4.5 kt of TNT. In contrast, a good sized thermonuclear weapon could deliver thousands of times that energy (even taking into account the relatively poor conversion of 100 megatons yield into delta-V).
I hope Office 12 has ways to turn off all the auto-fill, auto-format, auto-magically do-what-you-don't-want "features" that turn Office users into sobbing heaps. I've spent many an hour rooting around in Office Prefs (which for some reason you can only do when a document is open despite the fact that the prefs aren't document-specific?!?!?!) and have tried to lobotomize Office, but it keeps finding ways to auto-fsck my documents.
Office's "intelligent" features have a horrible accuracy rate for me, but then maybe I just think different.
I'd also request they fix all the bugs/annoyances that have lingered unfixed in Office 8, 9, 10, and 11 before they try to "enhance" Office any further.
Safety is a real problem. Lots of things kill cancer cells or have other useful medical properties. The problem is that too many of them also screw up other cells or bodily processes. Rapidly dividing cells, such as those in the digestive system and hair follicles are often hosed by cancer drugs (hence the nausea and hairlessness of cancer treatment patients). I'd also worry that anything that accelerates immune function leaves the patient prone to autoimmune diseases. Sure, I'd rather not die of cancer, but if it means I get MS, rheumatoid arthritis, diabetes, hypothyroidism, etc., then I (and the FDA) might think twice about approving the drug or really calling it "a cure."
Only about 1 in 5000 drugs gets approved (and I'd bet that all 4999 rejects started out promising). And only 1 in 3 approved drugs makes money for the company (i.e., is actually used to treat the number of patients that the company expected).
I truly wish them the best of luck in finding new treatments for cancer, but I also recognize that the odds are stacked against the new drugs.
Slashdot Mirror
Cisco has about $5 billion in cash & short-term investments. Nokia's market cap is $71 billion (and would probably need an offer that values the firm at $80 to $90 billlion to succeed). Even if Cisco liquidates its $11 billion in long-term investments, it can't swing this deal with what it has on hand.
Unless Cisco goes into major debt with a leveraged buyout, they can't afford to buy someone as big as Nokia.
What about creating an ad hoc distributed network of sensors (versus a fixed network). If thousands or millions of people downloaded a worm monitor application, then the sensor network would be very fluid and span IP space in a less predictable way. An ongoing P2P cross-comparison of the signatures of unsolicited packets could also provide distributed detection of novel worms. When too many sensors see the same anomalous thing, the alert propagates across the network.
Done well, it would create an internet immune system in that sensors that had seen the worm would alert machines that hadn't seen the worm. Those machines would then automatically filter for the new pattern while watching for confirming evidence that a worm was loose. If the download also provided a protective feature, then more people would download it and that the network would become more sensitive and valuable.
The idea may have some minor problems. First, security vulnerabilities could be introduced by the monitoring package itself (e.g, the Witty worm targeted vulnerable firewall software). I'd recommend that if a buffer overflow or malformed command exploit were ever discovered in Honeypot@home code, then all the developers would have to be shot immediately. Second, I see some but no great problem of worm writers trying to subvert the network because it would be hard for them to register enough machines (and replace the code to mute the alert signal triggered by their worm) to swamp the alert signals generated by legitimate nodes of the network.
I'm sure someone is working on this very thing.
There is a large assumption here, which is that only the customer who was issued the card is the person that ever uses the card, which is not true. Especially when said cards have incentives like "points" that can be accumulated to obtain free items, or a straight discount for using the card. Such card "borrowing" will skew the data in the database, since the computer assumes that the card equals the person, which is not necessarily true.
Card borrowing does skew the data, but how high is the actual frequency of this practice? Moreover, I'd bet that many card borrowings are between demographically similar people. If one 22 year old college student borrows a card from another 22 year old college student, then the effect on Tesco is minimal. Yes, card borrowing affects the data, but I doubt it ruins it.
You could argue that Tesco's buying policy really is only concerned with the aggregate information anyway since they won't make a purchase decision based on a single cardholder, and this aggregate is unchanged by card "swapping". But then why do you need to track individual information?
Aggregate data only gets a company so far. Consider the problem deciding where to put a new store. The only way Tesco can accurately do this is by cross comparing the census records for the proposed area with Tesco's own data on who buys how much and from what distances at its existing stores. If Tesco knows that people drive 20 miles to reach a Tesco, then its less likely to put another Tesco within 10 miles of an existing one. And if Tesco sees certain demographic groups is a new area, they can estimate likely sales and the best product assortment by studying the sales habits of those demographic groups within its loyalty card program. Breaking down the customer base by distance to store, types of goods bought, ages, income, etc. all help the company do a much better job of store siting, merchandise assortment, forecasting, etc.
Yes, only aggregate data is useful. But the loyalty card program lets Tesco aggregate data in new ways -- aggregate by customer age, gender, distance to store, household income, or tendencies to buy products. Store-level aggregate data or even simple market basket analysis isn't going to give as good an insight into the relationships between customer categories and shopping habits.
The answer is because they can. However the whole concept of "keeping a file" on someone is rather macabre. I have access to my criminal record, my medical record and my credit record. I should have access to all information ANYONE keeps on me unless it's being used to prosecute me for a crime. When is this going to happen at the supermarket? Until then, I refuse to use these cards.
I very seriously doubt that Tesco spends millions of pounds (probably tens of millions of pounds) on its loyalty card database just because it can. Most businesses insistent on a bottom-line improvement in the business before laying out that kind of money. And if a company does spend that type of money and doesn't get a profitable return, they stop the program. Morevoer, the type of data collected by a loyalty card program is perfect for assessing the business benefits of a loyalty card program. Tesco started their program 10 years ago. I doubt it would still be in use today if it didn't provide direct benefits to the company.
In essence, isn't the Tesco loyalty card system like a sophisticated representative democratic process? People "register to vote" using a loyalty card, vote by buying goods, and Tesco watches the results of the "election." Tesco knows who buys what and can thus go to suppliers and argue for changes that are more likely to satisfy customers.
Although non-loyalty card users still count at Tesco (all retail is a type of democracy in that people vote with their pocketbooks), I'm sure that the product choices of loyalty card users are far more influential with Tesco and thus with suppliers. In that regard not having a loyalty card is like not having a voter registration card.
Some might argue that voting should be anonymous, like political democratic elections, and perhaps it should. Yet non-anonymous voting provides valuable information -- e.g., Tesco might notice that, for example, people who buy lots of hot soups in the winter don't buy high-sugar fizzy drinks in the summer or some other correlation that is only observable if you can know that shopping basket A and basket B (6 months later) represent the same voter. These long-term correlations aid in both store assortment planning and forecasting.
It's easily the worst airport in the entire United States of America
Agreed! Trying to find a working electrical outlet in Concourse C (at least) is a true exercise in futility. Apparently MassPort disabled nearly all of them for some pathetic reason (concession to the for-pay internet kiosk maybe?).
GRRR! This thread reminded me that I have to fly there twice this month.
While I applaud attempts to secure WiFi, it would seem that wireless will always add another channel of vulnerability to any IT system, especially because WiFi is so often deployed inside the firewall. WiFi system are generally vulnerable to both internet-based attacks and wireless attacks. And even if the 802.11i protocol "secure," there is little guarantee that both the AP and the client wifi transceiver have a secure implementation of the protocol or that the user configures the system in a secure fashion.
As inconvenient as wires are (and even they are not totally secure), they do reduce the amount of one personal information freely broadcast into the ether.
You raise an excellent argument but I fear that it creates the opposite of what you want. Credit information is far more valuable (to humanity and civilization) than are the latest music files by Brittany Spears.
If you study banking in China you find that one of the big problems over there is a lack of credit information systems. Its easy for someone to get a loan, skip the payments, go get another loan at another bank, skip the payments, and repeat as needed. In such a system honest people pay the price (high interest rates) to cover losses generated by dishonest borrowers. Without some mechanism for sharing credit histories, its almost impossible to have a viable credit card system or low-cost consumer loans (I'll leave it to others to argue whether these are Good Things or not).
The problem, and it is a massive one, is not that people are collecting the information, but that they are misusing the information or allowing to be misused by failing to secure it against criminal incursions. The same aggressive defense that prevents counterfeit currency in the U.S. should be applied to those that would counterfeit identities with stolen information. Your point about Choicepoint is well taken -- collectors of personal financial information should be held very accountable (and liable) for lapses in their security and for the actions of those they give data to.
your "recomputing/rerendering the scene parallax in realtime" betrays your having forgotten that we're talking about film here...
Film records a fixed perspective but human heads and eyes are not fixed. And that's the problem with film for use in 3D. FIlm is not an appropriate medium for visually comfortable 3-D because it forces the viewer to hold their head in the same fixed orientation used by the cameras.
The core challenge for 3-D is creating a system that works when a person tilts their head. Current 3-D filming and multi-person viewing systems assume that the viewers left eye is a fixed horizontal distance to the left of the right eye with no vertical displacement between the eyes' pupils. This assumption is only true when everyone is sitting upright in their chairs. If the viewer tilts their head, then the parallax of the scene appears unnaturally displaced and gives the viewer eyestrain, headache, or a sensation of double-images. With 3-D, you can't rest your head on your partner's shoulder, tilt your head to see around the person in front of your, or lie on the couch and watch it without some visual discomfort. I'd imagine that most people won't consciously notice the problem but might subconsciously become aware that they get eyestrain, neck-pains, headaches, or a vaguely nauseous disoriented feeling when they see a 3D movie -- not a recipe for repeat business.
One nearterm solution to the problem is constructing tilt-dependent parallax for each viewer. The person with their head tilted to the right needs to see a different pair of images than the person who is sitting up straight or who has tilted their head to the left. This pushes 3D into the realm of more awkward and more expensive personal viewing headsets and the need for tracking head tilt and recomputing/rerendering the scene parallax in realtime.
The longterm solution is holographic or volumetric systems that create/reconstruct an optical 3-D field. This solves the head tilt problem, although adds the minor cinematic problem that the people on the left side of the theatre may have an obstructed view (relative to the people in the center or right-side) if, for example, the main character's hand covers some important object from some angles.
There's an old story from the early neural net image recognition days that seems germane to this. A group of researcher were trying to train an artificial neural net to recognize military tanks that were partially hidden in forested scenes (this was the bad old Cold War days and spotting Soviet tanks in West German forests was the problem du jour). Pictures of natural forested scenes with and without tanks were used to train and test the system. It seemed to work very well on all the training and test data.
But when they tried the system on more images, it failed miserably. Further investigation revealed that, by accident, all of the "tank" pictures had been taken on cloudy days and all of the non-tank pictures had been taken on sunny days. The system had learned, and learned beautifully, how to recognize cloudy vs. sunny days.
The point is that the software was good enough to learn to recognize the difference between the two populations of images but that that difference wasn't the one intended by the people working on the system. In the same vein, I'm sure that Peekaboom will learn to distinguish between objects in images but whether it learns the actual object or just some incidental characteristic of that pocture of the object will require a very very good diversity of training pictures to avoid accidental, non-meaningful patterns in the image data.
I do wish them luck. Perhaps Peekaboom could create a distributed version of the training process in which others can both submit and help train on new objects/images. Letting others submit images and train the system would help diversify the training & testing data sets. Because some people will, no doubt, submit porn, I'm sure the system might become quite adept at recognizing the nether regions of the human body.
Technological anonymity does not address all the vulnerabilities of the system because the identity of the source is also stored in the mind of the journalist. In fact, I suspect that technological security is an iron padlock on a paper house -- the human factors/social engineering issues create some severe vulnerabilities in the system. If the government can threaten or actually imprison a journalist over sources, then encrypted HDs aren't going to be much of a defense. As long as the journalist can ID the source, the system is very vulnerable.
The problem is that journalists really can't use unverified sources -- it's to easy to be either wrong, manipulated, or both as Dan Rather can attest to.
The real trick would be to accomplish both credibility and anonymity. Somehow the journalist needs to both know the source well enough to be sure that they are credible and yet not know the source well enough to ID them. Then they can add whatever encryption/obfuscation they want because none of the vulnerable information lies in a human mind.
If the OSS community could convince AOL to add a Linux distro to their ubiquitous CDs, I'm sure it would reach a lot of people. AOL may not be blanketing the world with disks like they used to, but they are still everywhere (in USPS change-of-address packets, next to store cash registers, and in the occasional Sunday newspaper).
I've never used it (maybe it deserves to die) but I'm surprised IBM didn't spin-off OS/2 sales & support as a little services company (with an appropriate slice of the proceeds of the service contracts). If people want to use OS/2, why not sell it to them? If people need support for it, why not sell it to them?
I could understand a company killing a product that competes with its own more modern systems, but how do continued OS/2 sales hurt IBM more than orphaning some existing customers?
Absolutely true. Anyone who thinks that some $/share is cheap or expensive without understanding total valuation is not going to last long in the stock market. The bigger challenge is figuring out what a share buys in terms of future returns.
Actually, Yahoo's revenues are growing faster than Google's. Earnings growth is a total red herring in fast-growing companies because its the difference of costs and revenues. In a young company its easy for these to flucuate and for the difference to gyrate madly (a zillion percent earning growth may mean nothing is the baseline was near zero). EBITDA is even worse because it subtracts numbers such as taxes, depreciation, and amortization that do come out of the company when it has matured. Because YHOO and GOOG are in much the same industry, they should have similar margins once they stabilize. Yet GOOG price/sales ratio is nearly twice that of YHOO's (21.88 vs. 12.85) suggesting that GOOG's margins need to be twice that of YHOOs to make up the difference. And if you care about it, YHOO has 4 times the cashflow of GOOG (although this number is hard to interpret in a rapidly expanding companies).
You are very perceptive to estimate the total cap of shorted stock. But I'd be careful about reading too much into the short ratios because shorters consider more than just the long-term future value of the stock in shorting. GOOG is currently a daring. If it is in a bubble, then shorters will avoid it even if they know that GOOG is horribly overvalued. Short sellers have a fixed maximum gain (i.e., when the stock goes to 0) but unlimited potential loss (e.g., if the stock double, triples, etc. in value). Because bubbles have a nasty habit of being unpredictable -- nobody knows when GOOG will hit the wall -- shorters will avoid GOOG. In contrast, YHOO has been through the dot-bomb ringer and that will make bulls more cautious and the bears more agressive. Short sellers know that YHOO is less likely to be caught up in a prolonged mania phase. The point, both GOOG and YHOO may have
The other front-page story on Netscape highlights the promises and risks of high-flying internet companies. In this post I argue that Netscape fell because it was so easy to switch browsers, especially when getting a new computer.
I wonder if Google will be able to make itself sticky enough to survive any threats? Currently, Google doesn't really offer any intercompatibility advantages in the sense that a co-worker's use of Google does not influence my use of Google. And if I replace my PC, Google doesn't offer anything that encourages me to use Google on the new machine. (GMail is somewhat sticky, but is too independent of Google's core search to force people to stay with Google search)
In contrast, I can see how MS could offer more integrative search experience where people would use MS search tools because friends and coworkers use MS search tools. If my coworker's PC is indexed by MS, my old PC was indexed by MS, and my new PC comes with built-in local/global search tools, then I'd bet a large fraction of people will switch to MS search tool regardless of Google's marketshare. Even Google's ad-words placements on 3rd party sites could be threatened if nex-gen MS server includes integrated ad serving tools.
I hope that Google finds a way to encourage people to stay with Google even as they change PCs or interconnect with co-workers and friends. The current valuation of Google requires both high growth and low risk.
Netscape's rise and fall epitomizes the acceleration of the business cycle. The fact that anyone can download anything at low cost and the fact that most people replace their computers every 2 years means a new, small company can quickly grow its customer base. And those same tools meant that MS could, just as quickly distribute its own browser and quickly take Netscape's installbase from the company.
Low distribution costs and PC turnover means that marketshare leadership is not assailable under most conditions -- its too easy for people to replace old software, especially when they get a new computer. Only companies that have an interoperability hook that ties past, present, and future generations of software and systems together have any hope of retaining marketshare.
MS has tried, and succeeded, in creating that hook with IE in that many websites "work best" with Explorer and Windows-specific web functionality (VBscript, ActiveX, MS-extensions to javascript, etc.). To the extent that MS is forced, in the future, to embrace true open standards (not embrace-and-extent forks of those standards) then the OS and app maker will become vulnerable to rapid changes in marketshare.
It sounds like we need new directives in robots.txt to categorize what's permitted and what's not permitted regarding caching, archiving, display of conextual text around a search hit, time-to-live, copyright jurisdiction, etc.
If some countries want to implement asinine policies, at least sites should have the decency (and the means) to let global information services, such as Google, respect those policies.
Connectivity - global media, the internet -- have created a winner-take-all world that drives both the creators of studies and the reporters of studies toward hyperbole. If someone wants their 15 minutes of fame, they need to do (or appear to do) something spectacular. When attention is a scarce resource (because of an explosion of applications/demands for attention), then it drives people toward excessive behavior in crafting and reporting the results of studies.
At the same time, I wonder if the long tail efect means that an increasing number of once-obscure, high-quality studies are being discovered, read, and used by an increasing number of people. Those that do create unbiased studies may not get much popular press, but they do become more widely read due to Google.
Ultimately, we seem to be floating in a rising tide of both good and bad studies. Perhaps the ratio of studies is being biased toward the bad (winner take all) but the ratio of impressions -- the numbers of times that good studies have been accessed -- has actually improved due to long-tail effects.
If everyone switches to wideband, low-power, densely-coded, mesh-network transmissions, then I suspect that the Earth will become virtually invisible to extraterrestrials who try to use SETI-style, pattern-in-RF methods. With nobody broadcasting at high power on a simple-coded narrow-band carrier, the RF emissions of the planet will become indistinguishable from noise.
I wonder if each civilization goes through a short RF-detectability phase before they so densely pack the spectrum with so many emitters that they become invisible, too.
This development highlights the ongoing replacement of specialized, engineered devices with general purpose CPU + software. So many things (car's carburetors, motor speed controls, printers, appliance controls, radios, etc.) were formerly designed using mechanical and electrical circuits that implemented the needed functionality. Now they do it with a CPU such as an embedded controller and a bit of code in flash-RAM.
/. cliche, but I, for one, welcome our new software-embodied overlords.
The shift from hardware-embodied functionality to software-embodied functionality is very profound because of the differences in cost structures. The cost of complexity for software is far lower than the corresponding cost of complexity for hardware. The cost of manufacturing for software is lower than the cost of manufacturing for hardware. The cost of modifying or upgrading software is far lower than the cost of replacing or upgrading hardware. Products with software-embodied functionality can be designed at low cost, made in volume at lost cost and changed at will after sale. The result is greater variety and faster development of new products.
The effects go much farther than cost and variety. Perhaps the most interesting effect is that caused by Moore's law. Whereas mechanical and electronic system have not improved much in the last few decades, CPUs have. Creating software-embodied systems means that the device's performance can become slaved to Moore's law -- a software-embodied product gets a free performance boost with every doubling of transistor count/clockspeed.
I apologize for using this
The low-end has always, eventually, overtaken the high-end. Minicomputer became as powerful as mainframes, Workstations became as powerful as minicomputers, and PCs became as powerful as workstations. There will always be room for some high-end gear, but total marketshare for high-end stuff isn't growing and some of the high-end players have died.
The interesting question is when consumer electronics will replace PCs as the most ubiquitous computing device. They are not there yet, but I'd wager that next generation consoles could easily run most of the applications that most people currently use a PC for (web, email, office, and games). Of course some people will always seek out $2,000+ high-end PCs, but more and more people will opt for a $300-$500 unit like a game console (with disk) or something like a Apple's Mini.
I'm sure this is why MS has devoted so much money to XBox and Apple created the Mini. They learned the lesson (that DEC both epitomized but failed to learn) that the low-end rises and supplants the high-end.
An estimate of the orbital delta-v for Tempel /Deep Impact suggests a velocity change of only 1 cm/hour (I can't vouch for the math). Assuming we would need to nudge a threatening body by 1/2 the diameter of the Earth (from direct hit to grazing pass-by), we would need to know to hit a Tempel 1-sized body in advance by over 73,000 years. This type of mission would work 10 years in advance for much smaller bodies (say less than 350 m in diameter). Even these estimates assume a perfect strike by the deflecting deep impactor -- a margin of error or the need to push the object several Earth-diameters further reduces the potential for this method.
Kinetic energy is not the way to go. Deep Impact delivered only about 4.5 kt of TNT. In contrast, a good sized thermonuclear weapon could deliver thousands of times that energy (even taking into account the relatively poor conversion of 100 megatons yield into delta-V).
I've been getting "chocolate bar" spam ever since news came out that almost three quarters of office workers in an impromptu man-on-the-street survey were willing to give up their passwords when offered the bribe of a chocolate bar.. The spam claims to provide 10 pounds of Hershey's chocolate in exchange for who knows what.
I hope Office 12 has ways to turn off all the auto-fill, auto-format, auto-magically do-what-you-don't-want "features" that turn Office users into sobbing heaps. I've spent many an hour rooting around in Office Prefs (which for some reason you can only do when a document is open despite the fact that the prefs aren't document-specific?!?!?!) and have tried to lobotomize Office, but it keeps finding ways to auto-fsck my documents.
Office's "intelligent" features have a horrible accuracy rate for me, but then maybe I just think different.
I'd also request they fix all the bugs/annoyances that have lingered unfixed in Office 8, 9, 10, and 11 before they try to "enhance" Office any further.
Safety is a real problem. Lots of things kill cancer cells or have other useful medical properties. The problem is that too many of them also screw up other cells or bodily processes. Rapidly dividing cells, such as those in the digestive system and hair follicles are often hosed by cancer drugs (hence the nausea and hairlessness of cancer treatment patients). I'd also worry that anything that accelerates immune function leaves the patient prone to autoimmune diseases. Sure, I'd rather not die of cancer, but if it means I get MS, rheumatoid arthritis, diabetes, hypothyroidism, etc., then I (and the FDA) might think twice about approving the drug or really calling it "a cure."
Only about 1 in 5000 drugs gets approved (and I'd bet that all 4999 rejects started out promising). And only 1 in 3 approved drugs makes money for the company (i.e., is actually used to treat the number of patients that the company expected).
I truly wish them the best of luck in finding new treatments for cancer, but I also recognize that the odds are stacked against the new drugs.